Small Circular DNA Molecules Act as Rigid Motifs To Build DNANanotubesHongning Zheng, Minyu Xiao, Qin Yan, Yinzhou Ma, and Shou-Jun Xiao*

Nanjing National Laboratory of Microstructures, State Key Laboratory of Coordination Chemistry, School of Chemistry andChemical Engineering, Nanjing University, Nanjing 210093, Jiangsu, China

*S Supporting Information

ABSTRACT: Small circular DNA molecules with de-signed lengths, for example 64 and 96 nucleotides (nt),after hybridization with a few 32-nt staple strandsrespectively, can act as rigid motifs for the constructionof DNA nanotubes with excellent uniformity in ringdiameter. Unlike most native DNA nanotubes, whichconsist of longitudinal double helices, nanotubesassembled from circular DNAs are constructed fromlateral double helices. Of the five types of DNA nanotubesdesigned here, four are built by alternating two differentrings of the same ring size, while one is composed of all thesame 96-nt rings. Nanotubes constructed from the same96-nt rings are 10−100 times shorter than thoseconstructed from two different 96-nt rings, because thereare fewer hinge joints on the rings.

The mechanical properties of small circular DNAs (tens toa hundred or more base pairs), such as bending, twisting,

curving, kinking, stressing, rigidity, and disruption, haveattracted scientists’ great interest because they relate closelyto the secret of our lifemolecular mechanisms and functionsof all kinds of DNA topological structures.1 In addition, theirmolecular scale and geometry are well suited to mathematicaltopology model systems for quantitative analysis. From studiesof the well-known wormlike model, such circular double-helixDNAs have been generally accepted to behave as rigid ringswith smooth bending, i.e., small angular changes between theplanes of adjacent base pairs, in solution over decades.2 Fromanother side, DNA nanotechnology pioneered by N. D.Seeman is based on rigid DNA motifs as building blocks toconstruct zero-, one-, two-, and three-dimensional (abbreviatedas 0D, 1D, 2D, and 3D) nanostructures.3 Here we address howsmall circular DNAs can act as rigid motifs to constructnanotubes, which are not yet as well-recognized in DNAnanotechnology as those made from linear strands. Such DNAminicircle-based nano-architecture might provide a newperspective for understanding the mechanical properties ofcircular DNAs.In the early development stages, Seeman’s group adopted a

post-ligation strategy to close the nicking sites of polyhedralobjects including cubes,4 truncated octahedra,5 and Borromeanrings,6 assembled from linear strands, which stabilized theflexible nanostructures when the DNAs were in their linear andend-open states. Similarly, Turberfield et al. constructed shape-persistent objects from DNA tetrahedra with high diaster-

eoselectivity by post-ligation.7 Another strategy employingsmall circular DNAs as rigid building blocks for DNA nano-architectures uses pre-circularized and purified DNAs directlyin the assembly. Mao and co-workers used either circular orlinear DNAs possessing the same sequence, for structuralsymmetry, to construct three- and four-arm building blocks andthen hundreds of nanometer or micrometer 2D DNA networkarrays.8 Recently they also reported the use of circularized DNAas a “lid” to construct nanotubes with defined diameters andlengths.9 In their designs, either circular or linear strands withthe same sequence can replace each other without causingsignificant differences in the assembled nanostructuralmorphology. The explanation for this is that the core motifsin those designs are rigid tiles of double crossovers, and pairs ofneighboring rigid tiles were connected with 4−9 thyminesserving as hinges between them. Famulok et al. used DNAminicircles with gaps to construct rotaxane-like nano-robotsand other nano-devices.10 Sleiman et al. developed artificialDNA circles as motifs by inserting three or four rigid organicspecies to modify a small circular DNA into a triangle or asquare, which then formed so-called lateral “rungs” that werelinked longitudinally by double or single strands to constructDNA nanotubes.11 The design benefits from the rigid lateralrungs consisting of rigid organic vertices and double-helixsegments. Such chemically modified DNA strands could havepotential applications in DNA self-assembly; however, theyrequire laborious synthesis and purification, and again theartificial DNAs possess very different structures, making theirassembly strategy somewhat different from that of nativestrands: e.g., the neighboring hinge joints “up” and “down” onlateral rungs were separated by a distance of two turns (21 basepairs (bp)), while the distance between two neighboringcrossovers for origami built from native strands is 1.5 or 2.5turns.Herein, we report a new type of nanotubes assembled from

native small circular DNAs, taking the advantage of thepredicted rigidity of double-helix rings in solution. Two kindsof DNA single-stranded minicircles, with lengths of 96 and 64nucleotides (nt), were used to demonstrate the feasibility ofconstructing geometrically well-defined DNA nanotubes ofexcellent uniformity from a few 32-nt staple strands withoutleaving a single base unpaired. Different from most previouslyreported native DNA nanotubes composed of longitudinaldouble helices, our nanotubes, assembled from circular DNAs,

Received: April 23, 2014

Communication

pubs.acs.org/JACS

© XXXX American Chemical Society A dx.doi.org/10.1021/ja504050r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

are composed of lateral double helices. In all five of our designs(NT1−NT5), the circular single strands have bending, twisting,and kinking flexibility for hybridization with staple strands, andonce these rings become completely double-helixed, theybecome rigid and can act as building blocks in DNAnanotechnology. The 32-nt complementary staple strandsspanning two or three neighboring rings serve partly asbuilding blocks when hybridized and partly as hinge jointswhen crossed over two or three neighboring rings to connect allthe rings together. A vivid analogy for this kind of nanotubes isa tower constructed of kiss-stacked storeys.The general approach to prepare small circular DNAs is

enzymatic ligation. We applied T4 DNA ligase to close thenicking site with the help of a splint oligonucleotide(Supporting Information, Figure S1) and then purified thecircularized DNAs by PAGE (Figures S2−S4).Our first and second nanotube designs, NT1 and NT2, are

depicted in Figure 1. The complete folding process for NT1

(Figure 1c) is illustrated in Figure 1a−d: two different 96-ntrings (C1 and C2, Figure 1a) are kiss-stacked anti-parallel toeach other along the central axis, alternating six 32-nt staplestrands (Figure 1b) to build NT1 (a vivid 3D illustration isshown in Figure 1c), a longitudinal section of which, with abreakline along the direction of the arrow shown in Figure 1c, isdemonstrated in Figure 1d. Each 32-nt staple is divided intothree concatenated parts8 nt +16 nt + 8 ntspanning threeneighboring rings, and the long 16-nt sections from six staplesare placed alternately on C1 and C2. According to thecrossover rule in origami, the distance between twoneighboring crossovers with opposite directions is 1.5 turns(or 16 bp); therefore, each ring bears six crossovers (96/16 =6), which are designed to alternate in up and down directions.

For NT2, the long 16-nt sections from six staples are placedon the same ring (C3, Figure 1e). The complete foldingprocess for NT2 (Figure 1g) is illustrated in Figure 1e−h. Thefolding strategy adopted here is similar to our previousreports12 on folding rolling circle amplified DNAs or rollingcircle transcripted RNAs with a few staple strands, except that

in this case small circular DNA molecules replace the longsingle scaffolding DNA or RNA strands, and the resultingassembled nanostructures are nanotubes instead of nanorib-bons.Analyses of nanotubes NT1 and NT2 using atomic force

microscopy (AFM) are demonstrated in Figure 2. Figure 2a−c

reveals the well-defined DNA nanotubes NT1 from lower tohigher magnification, with image sizes at 12 × 12, 6 × 6, and 3× 3 μm2, respectively, while Figure 2e−g presents NT2 at 14 ×14, 6 × 6, and 3 × 3 μm2, respectively. Although the molecularstructures of NT1 and NT2 are designed somewhat differently,their nanoscale morphologies in AFM look the same. Thetypical lengths of NT1 and NT2 are 3−15 μm. The cross-section analyses (Figure 2d,h) show a height of 2.8 nm and awidth of 15 nm. Theoretically, the diameter of the tube can beeasily calculated from the formula d = C/π = (96 × 0.34 nm)/π≈ 10 nm for the nanotubes in the B-form duplex, where Crepresents the perimeter of the double-helix ring (96 × 0.34 nm≈ 32 nm), 96 being the number of base pairs of the ring and0.34 nm the length per base pair. Obviously, the measuredheight and width are not the parameters to describe such ananotube. We executed AFM measurements in air within 8 hafter DNA’s adsorption on mica, during which time thenanotube lost water molecules and was squashed to a doublelayer. This squashed double layer has a width of around C/2 =16 nm and a height of 2.8 nm (twice that of a double-helixDNA, 1.4 nm), corresponding very well to the measureddimensions.Can we use only a single type of DNA ring molecules to cost-

efficiently construct nanotubes? Figure 3 presents this as ourthird design: the same 96-nt rings (C1 in Figure 3a) are kiss-stacked anti-parallel to each other along the central axis bythree 32-nt staple strands (Figure 3b) to build NT3 (Figure3c). This kind of nanotubes could be constructed following thecrossover rule of origami, with alternating crossovers at adistance of 1.5 turns (see Figure S7). However, according tothe DNA strand displacement rule,13 we suggest the foldingstructure of NT3 shown in Figure 3c as a vivid 3D view and inFigure 3d with a longitudinal section. Obviously only the

Figure 1. Folding strategies for the formation of nanotubes NT1 andNT2 from two 96-nt rings and six 32-nt complementary staple strands.(a) Cartoons of two 96-nt rings, with C1 in blue and C2 in black. (b)Six 32-nt staple strands shown in different colors. (c) Vivid 3Dillustration of NT1. (d) Longitudinal section of NT1 with a breaklinealong the direction of the arrow in (c). (e) Cartoons of two 96-ntrings, C1 in blue and C3 in brown. (f) Six 32-nt staple strands shownin different colors. (g) Vivid 3D illustration of NT2. (h) Longitudinalsection of NT2 with a breakline along the direction of the arrow in (g).

Figure 2. AFM characterization of nanotubes NT1 and NT2. (a−c)AFM images of NT1 from lower to higher magnification with imagesizes at 12 × 12, 6 × 6, and 3 × 3 μm2, where (b) is a zoom-in view of(a) and (c) is a zoom-in view of (b). (d) Cross-section profile of thedashed line in (c). (e−g) AFM images of NT2 from lower to highermagnification with image sizes at 14 × 14, 6 × 6, and 3 × 3 μm2, where(f) is a zoom-in view of (e) and (g) is a zoom-in view of (f). (h)Cross-section profile of the dashed line in (g). Both (d) and (g) showthe same height of 2.8 nm and the same width of 15 nm for NT1 andNT2.

Journal of the American Chemical Society Communication

dx.doi.org/10.1021/ja504050r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXXB

orange 32-nt staple strand spans three neighboring rings, whilethe other two 32-nt staple strands (green and red) span twoneighboring rings and form two 32-nt rectangles. In this case,each ring holds only four crossovers, which can be classifiedinto two groups. Each group contains one up crossover and onedown crossover at both hinge joints of the orange staples, andthe two groups are separated by a green rectangle at one sideand a red one at the other side.AFM analyses in Figure 3e−g illustrate the typical length of

nanotubes NT3 to be 100−500 nm, 10−100 times shorter thanNT1 and NT2. A lot of nanotube fragments with length lessthan 100 nm are also observed. The cross-section profile inFigure 3h presents a height of 2.8 nm, i.e., a squashed bilayer ofNT3, and a width of 15 nm, i.e., half the perimeter of NT3,which are the same as those of NT1 and NT2. The muchshorter length of nanotubes NT3 proves the correct foldingstructure in Figure 3c,d because they have two fewer hingejoints on each ring than NT1 and NT2 and thus weakerbinding to connect the rings together. To elaborate on thecorrect folding structure, we can look at it from the reverseside: according to the symmetry rule suggested by Mao,9 if thenanotubes are folded as in Figure S7, they will be much longerthan NT1 and NT2.Next, we changed the ring size from 96 to 64 nt. Figure 4

illustrates the strategies for assembly of two 64-nt rings intonanotubes. Since the 64-nt ring can hold only four crossovers(64/16 = 4), two up and two down, only four 32-nt staplestrands are needed for the assembly. Similar to NT1 and NT2,two different 64-nt rings (C4 and C5 in Figure 4a) are kiss-stacked anti-parallel to each other along the central axis,alternating four 32-nt staple strands (Figure 4b,e) to form NT4(Figure 4c) and NT5 (Figure 4f). Figure 4d,g shows thecorresponding longitudinal sections with a breakline along thearrow direction in NT4 or NT5, respectively. The onlydifference between NT4 and NT5 is that in NT5 all four longsections (16 nt) from the four 32-nt staples are placed on C5,while in NT4 the four long 16-nt sections from four 32-ntstaples are allotted equally to C4 and C5.AFM analyses in Figure 5 illustrate that the typical length of

nanotubes NT4 and NT5 to be 0.5−3 μm. Cross-sectionanalyses in Figure 5d,h show a height of 2.3 nm, i.e., a squashedbilayer of NT4 or NT5, and a width at 10 nm, i.e., half the

perimeter of NT4 or NT5 [(64 × 0.34 nm)/2 = 11 nm],regardless of the type of folding. The diameters of NT4 andNT5 can be easily calculated as (64 × 0.34 nm)/π ≈ 7 nm.Can only a single type of 64-nt rings with two 32-nt staples

be used to construct nanotubes, similar to NT3? According tothe strand displacement rule, a single type of 64-nt rings moreeasily forms dimers with two anti-parallel rings instead ofnanotubes with many interconnected rings.As we expected, the nanotubes built from DNA minicircles

are unique because they are constructed completely from lateraldouble helices only. The production yield is very high. Someunique characteristics of these nanotubes are emphasized: theirdiameter is geometrically well-defined with excellent uni-formity; there are no open nanotubes with saw-tooth edges

Figure 3. Construction of nanotube NT3 from C1 and 3 × 32 staplestrands. (a) Cartoon of a single 96-nt ring of C1 in blue. (b) Three 32-nt staple strands shown in different colors. (c) Vivid 3D illustration ofNT3. (d) Longitudinal section of NT3 with a breakline along thedirection of the arrow in (c). (e−g) AFM images of NT3 from lowerto higher magnification with image sizes at 6 × 6, 3 × 3, and 1 × 1μm2, respectively. (h) Cross-section profile of the dashed line in (g).

Figure 4. Folding strategies for the formation of nanotubes NT4 andNT5 from two 64-nt rings and four 32-nt complementary staplestrands. (a) Cartoons of two 64-nt rings, with C4 in blue and C5 inblack. (b) Four 32-nt staple strands shown in different colors. (c) Vivid3D illustration of NT4. (d) Longitudinal section of NT4 with abreakline along the direction of the arrow in (c). (e) Four 32-nt staplestrands shown in different colors. (f) Vivid 3D illustration of NT5. (h)Longitudinal section of NT5 with a breakline along the direction ofthe arrow in (g).

Figure 5. AFM characterization of nanotubes NT4 and NT5. (a−c)AFM images of NT4 from lower to higher magnification with imagesize at 9 × 9, 6 × 6, and 3 × 3 μm2, respectively, where (b) is a zoom-in view of (a) and (c) is a zoom-in view of (b). (d) Cross-sectionprofile of the dashed line in (c). (e−g) AFM images of NT5 fromlower to higher magnification with image size at 9 × 9, 6 × 6, and 3 ×3 μm2, respectively, where (f) is a zoom-in view of (e) and (g) is azoom-in view of (f). (h) Cross-section profile of the dashed line in (g).Both cross-section profiles show the same height of 2.3 nm and thesame width of 10 nm for NT4 and NT5.

Journal of the American Chemical Society Communication

dx.doi.org/10.1021/ja504050r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXXC

as are often observed in nanotubes built from longitudinaldouble helices;3h circular DNAs are less degradable than linearDNAs in a physiological environment;1l and they are wellmono-dispersed but seldom aggregated together in solution.These unique properties might be preferable in biologicaltechnologies such as drug/gene delivery, gene up- and down-regulation, gene expression, and disease diagnosis andtherapeutics.14

The successful assembly of nanotubes from small circularDNAs illustrates that the double-helix minicircles possessexcellent rigidity for DNA nano-architectures. Meanwhile, thecrossovers as hinge joints presents some bending flexibility andrestricted torsional freedom in linking these small ringstogether. Both rigidity and bending flexibility need to beunited to assemble the DNA minicircles into nanotubes. Thiswork partly proves the power of predictions based onquantitative calculation of the topology from the experimentalside. With further investigations, DNA minicircle-based nano-technology will present a prosperous future.

■ ASSOCIATED CONTENT*S Supporting InformationExperimental procedures, PAGE bands, sequence codes, designdetails, and more AFM images. This material is available free ofcharge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Authorsjxiao@nju.edu.cnNotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe are thankful for the financial support of NSFC, No.91027019, the National Basic Research Program of China, No.2013CB922101, and funding from the State Key Laboratory ofBioelectronics of Southeast University. Ziqi Lin is appreciatedfor providing the vivid 3D drawings.

■ REFERENCES(1) For example, see: (a) Lederberg, J. Physiol. Rev. 1952, 32, 403.(b) Mathematics of DNA Structure, Function and Interactions; Benham,C. J., Harvey, S., Olson, W. K., Sumners, D. W., Swigon, D., Eds.; IMAVolumes in Mathematics and Its Applications 150; Springer: TwinCities, 2009. (c) Vologodskii, A. V. Circular DNA. Education Portal ofthe Biophysical Society, 1999; http://www.biophysics.org/portals/1/pdfs/education/vologodskii.pdf. (d) Minicircle and Miniplasmid DNAVectors; Schleef, M., Ed.; Wiley-VCH: Weinheim, 2013. (e) Shore, D.;Langowski, J.; Baldwin, R. L. Proc. Natl. Acad. Sci. U.S.A. 1981, 78,4833. (f) Shore, D.; Baldwin, R. L. J. Mol. Biol. 1983, 170, 957.(g) Crick, F. H.; Klug, A. Nature 1975, 255, 530. (h) Vologodskii, A.;Du, Q.; Frank-Kamenetskii, M. D. Artif. DNA: PNA XNA 2013, 4, 1.(i) Vafabakhsh, R.; Ha, T. Science 2012, 337, 1097. (j) Vologodskii, A.;Frank-Kamenetskii, M. D. J. Am. Chem. Soc. 2013, 41, 6785.(k) Zheng, X.; Vologodskii, A. Biophys. J. 2009, 96, 1341. (l) Du,Q.; Kotlyar, A.; Vologodskii, A. Nucleic Acids Res. 2008, 36, 1120.(m) Timothy, E.; Cloutier, T. E.; Widom, J. Mol. Cell 2004, 14, 355.(n) Du, Q.; Smith, C.; Shiffeldrim, N.; Vologodskaia, M.; Vologodskii,A. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 5397.(2) For example, see: (a) Shimada, J.; Yamakawa, H. Macromolecules1984, 17, 689. (b) Zhang, Y.; Crothers, D. M. Biophys. J. 2003, 84,136.(3) For example, see: (a) Seeman, N. C. Nature 2003, 421, 33.(b) Winfree, E.; Liu, F.; Wenzler, L. A.; Seeman, N. Nature 1998, 394,

539. (c) Rothemund, P. Nature 2006, 440, 297. (d) He, Y.; Ye, T.; Su,M.; Zhang, C.; Ribbe, A. E.; Jiang, W.; Mao, C. Nature 2008, 452, 198.(e) Douglas, S. M.; Dietz, H.; Liedl, T.; Hogberg, B.; Graf, F.; Shih, W.M. Nature 2009, 459, 414. (f) Zheng, J.; Birktoft, J. J.; Chen, Y.; Wang,T.; Sha, R.; Constantinou, P. E.; Ginell, S. L.; Mao, C.; Seeman, N. C.Nature 2009, 459, 414. (g) Han, D.; Pal, S.; Nangreave, J.; Deng, Z.;Liu, Y.; Yan, H. Science 2011, 332, 342. (h) Yin, P.; Hariadi, R. F.;Sahu, S.; Choi, H. M. T.; Park, S. H.; LaBean, T. H.; Reif, J. F. Science2008, 321, 824. (i) Iinuma, R.; Ke, Y.; Jungmann, R.; Schlichthaerle,T.; Woehrstein, J. B.; Yin, P. Science 2014, 344, 65.(4) Chen, J. H.; Seeman, N. C. Nature 1991, 350, 631.(5) Zhang, Y. W.; Seeman, N. C. J. Am. Chem. Soc. 1994, 116, 1661.(6) Mao, C. D.; Sun, W. Q.; Seeman, N. C. Nature 1997, 386, 137.(7) Goodman, R. P.; Schaap, I. A. T.; Tardin, C. F. C.; Erben, M.;Berry, R. M.; Schmidt, C. F.; Turberfield, A. J. Science 2005, 310, 1661.(8) He, Y.; Tian, Y.; Chen, Y.; Deng, Z.; Ribbe, A. E.; Mao, C. Angew.Chem. 2005, 117, 6852.(9) Qian, H.; Tian, C.; Yu, J.; Guo, F.; Zheng, M.; Jiang, W.; Dong,Q.; Mao, C. Small 2013, 10, 855.(10) Ackermann, D.; Schmidt, T. L.; Hannam, J. S.; Purohit, C. S.;Heckel, A.; Famulok, M. Nat. Nanotechnol. 2010, 5, 436.(11) Aldaye, F. A.; Lo, P. K.; Karam, P.; McLaughlin, C. K.; Cosa, G.;Sleiman, H. F. Nat. Nanotechnol. 2009, 72, 349.(12) (a) Ma, Y.; Zheng, H.; Wang, C.; Yan, Q.; Chao, J.; Fan, C.;Xiao, S. J. J. Am. Chem. Soc. 2013, 135, 2959. (b) Zheng, H.; Ma, Y.;Xiao, S. J. Chem. Commun. 2014, 50, 2100.(13) Zhang, D. Y.; Seelig, G. Nature Chem. 2011, 3, 103.(14) Lu, C. H.; Willner, B.; Willner, I. ACS Nano 2013, 7, 8320.

Journal of the American Chemical Society Communication

dx.doi.org/10.1021/ja504050r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXXD